42 Proceedings of the Royal Society of Edinburgh. [Sess. 
which is set so that the middle of the spectrum suffers minimum deviation,, 
falls on the mirror N, and is brought to a focus on the photographic plate. 
The mirror N is rotated about a vertical axis through its centre by means 
of a micrometer screw ; hence all the colours of the spectrum can be made 
to pass in succession across the centre of the photographic plate. The 
curve along which the spectrum lies can be calculated from the formula 
for the oblique incidence of a thin pencil on a concave mirror. It does 
not deviate appreciably from a straight line. 
The concave mirrors had a diameter of 5 cm. and a focal length of 
333 cm. The sides of the quartz prism were 4 cm. high and 4 6 cm. 
long, and the two halves were held together in a brass frame. 
For the material of the mirrors I have taken nickel. Professors Hagen 
and Rubens have determined the reflecting power of various metals 
throughout the ultra-violet, visible, and infra-red, and the results show (cf. 
Landolt and Bornstein’s Tables) that silver reflects one region in the ultra- 
violet very poorly while reflecting excellently in the visible and infra-red. 
Spiegel-magnalium reflects well throughout the whole spectrum but does 
not keep, and is consequently not used in optics now. We tried the alloy , 
68 per cent, copper, 32 per cent tin, but rejected it on account of its 
brittleness. The first nickel mirrors obtained were silver on which nickel 
was deposited electrolytically. They were unsuitable, as was mentioned in 
the former paper. Then solid mirrors of 99 per cent, nickel were pro- 
cured, but their polish was not good enough. Finally, a purer quality of 
nickel was obtained, and the mirrors made from it were satisfactory in 
every particular. 
In the most unfavourable part of the ultra-violet nickel reflects 37 *8 
per cent, of the incident light. If we allow for the double reflection, only 
14'3 per cent, of the incident light is finally reflected. The fraction trans- 
mitted by two quartz lenses in succession in the ordinary spectrograph 
cannot be greater than 80 per cent. But owing to the obliquity of the 
plate we have to multiply this by sin 21° in order to compare correctly the 
energy received per unit area of the photographic plate in the two cases. 
This gives 28 per cent. Hence, when quartz lenses are used, and the 
aperture of the instrument is the same, the illumination of the image on 
the photographic plate cannot be more than twice as great as when nickel 
mirrors are used. With nickel mirrors, owing to the better collimation, 
a larger aperture can be used ; also I have a pair of silver mirrors for re- 
placing the nickel mirrors for use in the visible spectrum. When all things 
are considered, I am therefore of opinion that the mirror spectrograph gives 
the brighter spectra. It has, in addition, the following advantages : — 
